a defense of the traditional war convention* · 2015. 2. 22. · no less controversial) conception...

Ethics 118 (April 2008): 464–495� 2008 by The University of Chicago. All rights reserved. 0014-1704/2008/11803-0005$10.00

464

A Defense of the Traditional WarConvention*

Yitzhak Benbaji

The language of self-defense is used both in justifying wars and in ar-ticulating what is right and wrong within wars. In particular, the tradi-tional war ethic, especially as Michael Walzer formulates it in his Justand Unjust Wars, strongly suggests that a just war is a large-scale exerciseof the right of self-defense.1 Yet, critics of this traditional doctrine insistthat, in fact, the war convention reflects a particularly repugnant con-ception of the right of self-defense: “The just war theory, as manifestedin the war convention, seems much more like an adaptation to thecircumstances of war of a Hobbesian vision of the right of self-defensecombined with various elements of chivalric morality.”2

In order to understand the basis of this critique, let me sketch theHobbesian ethics of self-defense. Somewhat metaphorically, its bottomline can be put as follows: if a person’s life is at stake, “the notions ofRight and Wrong, Justice and Injustice have . . . no place.”3 The ar-gument contains two stages. First, in the state of nature, a person has

* I am grateful to Shlomo Avineri, Hagit Benbaji, Asa Carlson, Andrew Chignell,Tsilly Dagan, George Fletcher, Paul Gilbert, Judith Lichtenberg, David Luban, MikeLeBuffe, Eric Maskin, Avishai Margalit, Peter Pagin, Jennifer Pitts, Ron Shapira, AndrewChignell, Assaf Sharon, Levi Spectre, Torbjorn Tannsjo, David Widerker, Dror Yinon, andNoam Zohar for helpful conversations. I benefited enormously from written commentsby Gerhard Øverland, Ira Schnall, Danny Statman, and the editors of Ethics. Michael Walzerpatiently read several versions of this article; I am deeply thankful to him for his friendshipand advice. I am especially indebted to Jeff McMahan for his generosity and constantencouragement and for his close reading and extensive comments.

1. Michael Walzer, Just and Unjust Wars (New York: Basic, 1977), 61, 145. See also hisArguing about War (New Haven, CT: Yale University Press, 2004). Following Jeff McMahan,whose work is my main focus in this article, I treat the theory developed in Just and UnjustWars as the traditional just war doctrine or the traditional war convention.

2. Jeff McMahan, “Killing in War: A Reply to Walzer,” Philosophia 34 (2006): 47–51,51.

3. Thomas Hobbes, Leviathan (London: Crooke, 1651), 79.

Benbaji Defense of Traditional War Convention 465

a right—Hohfeldian liberty4—to do whatever he thinks is necessary forhis survival; that is, individuals possess no natural precontractual rightclaim to their own life. Second, in no contractual context can a personwaive his right liberty to harm others—or even to kill them—if this isnecessary for his survival. “A covenant not to defend myself from forceby force is always void,”5 as, for Hobbes, surrender of rights can bevalidated by considerations of self-interest (reason) only. And, it can beknown a priori that if it is determined that either a victim or a bystanderwould survive, but not both, no mutually beneficial contract betweenthem is possible. So, according to the Hobbesian ethics of self-defense,you violate no duty if you shield yourself from the bullets that are aboutto hit you by using a bystander who just happened to be around (incase this is necessary for your survival).

Hobbes’s ethics of self-defense (in general) and his denial of anatural right claim to life (in particular) are widely dismissed. Indeed,following Locke, most ethicists take the existence of a natural right claimto life to be a self-evident moral fact. Specifically, X has a natural rightclaim against Y that Y won’t kill him, even if the killing of X is necessaryfor Y’s survival. X might lose this claim against Y’s defensive attack inone case: if he poses a threat to Y’s life while having no right to do so.But, by definition, innocent bystanders have done nothing that couldhave caused them to lose their natural right to life.

To repeat, then, Walzer’s critics argue that the war convention, asit is explicated, systemized, and defended in Just and Unjust Wars, em-braces a Hobbesian conception of the right of self-defense. I shall focuson two targets of Jeff McMahan’s accurate and powerful articulation ofthis critique.6 The first is the traditional doctrine of the moral equalityof soldiers, according to which, within a war, just combatants (thosewho participate in a defensive war) and unjust combatants (those whoparticipate in a war of aggression) have an equal right to kill each otherin self-defense. Second, in the traditional view, although war does noteliminate morality, it “collectivizes” it: “materially” innocent people—those who do not pose any threat (i.e., a substantial risk of causingserious harm) to others—become liable to defensive killing, merely byvirtue of their membership in the military forces.

4. I follow Judith Thomson, The Realm of Rights (Cambridge, MA: Harvard UniversityPress, 1990), 49–50.

5. Hobbes, Leviathan, 86.6. See Jeff McMahan, “The Ethics of Killing in War,” Ethics 114 (2004): 693–733; cf.

his “War and Self-Defense,” Ethics and International Affairs 18 (2004): 75–80. For an earlierstatement, see his “Innocence, Self-Defense, and Killing in War,” Journal of Political Philos-ophy 2 (1994): 193–221. For related criticism, see David Rodin, War and Self-Defense (Oxford:Clarendon, 2002), and David Mapel, “Coerced Moral Agents? Individual Responsibilityfor Military Service,” Journal of Political Philosophy 6 (1998): 171–89.

466 Ethics April 2008

Both doctrines—McMahan argues—allow violating the right to lifeof innocent people. After all, just combatants use force in order toeliminate the unjust threats posed to their country by an aggressivestate; just combatants exercise their right to self- and other defense.Additionally, nonthreatening soldiers pose no threat to others. That is,just combatants and nonthreatening soldiers have done nothing thatcould have caused them to lose their right to life.

This article offers a defense of the moral equality of soldiers andthe collectivized morality of the war convention. I shall show that, despiteMcMahan’s claim to the contrary, rather than constituted by the Hobbes-ian (and thus excessively permissive) conception of self-defense, thetraditional war convention reflects a Lockean (i.e., extremely con-strained) conception of the scope of this right. My aim here is limitedin one central respect. I do not argue for the controversial (Lockean)conception of self-defense that I shall use in elucidating the traditionalwar convention. I take it for granted. Also, I shall not try to refute (theno less controversial) conception of self-defense that underlies Mc-Mahan’s unorthodox war ethic. (I undertook both tasks elsewhere.)7

Rather, by dissolving the deeply disturbing worries raised by his critique,I show that the traditional war convention does not collapse to theHobbesian conception of self-defense, despite its commitments to themoral equality of soldiers and to collectivizing their moral status.

I. THE TRADITIONAL WAR CONVENTION AND ITS CRITIQUE

A. Thomson’s Theory of Self-Defense and the Traditional War Convention

As noted, any theory of self-defense should explain why it is typicallypermissible to kill an attacker in self-defense, while it is impermissibleto kill a bystander in self-preservation. The moral requirement to dis-criminate attackers from bystanders is basic and uncontroversial. Not-withstanding this broad agreement, theorists formulate and explain thisrequirement in radically different ways.

Judith Thomson’s theory of self-defense provides a simple and at-tractive formulation of the basic requirement.8 Self-defensive killing ismorally justified, she argues, on the basis of facts about rights violation:the potential victim has a claim against the attacker that the attackerwill not kill him or pose an imminent threat to his life. Since these

7. In Yitzhak Benbaji, “Culpable Bystanders, Innocent Threats, and the Ethics of Self-Defense,” Canadian Journal of Philosophy 35 (2005): 623–40, I developed the theory of self-defense that I shall employ here. In Yitzhak Benbaji, “The Responsibility of Soldiers andthe Ethics of Killing in War,” Philosophical Quarterly 57 (2007): 558–73, I show that Mc-Mahan’s justice-based conception of self-defense yields a nearly pacifistic ethics of killingin war.

8. Judith J. Thomson, “Self-Defense,” Philosophy & Public Affairs 20 (1991): 283–310.


claims are Hohfeldian, the attacker is subject to the correlative duties.In a case in which the attacker violated these duties, the victim mightacquire a right to kill the attacker in self-defense. As opposed to thetypical attacker, a bystander violates no right—hence, he is not liableto killing.

Famously, Thomson believes that rights violation is possible withoutculpability, responsibility, or even agency. Thus, consider her “InnocentThreat” case:

Innocent Threat: A victim is sitting on a bench when he realizes thata man is falling on him. The falling man was pushed by a villainousaggressor, so he constitutes a threat without doing anything. Unlessthe victim blows the falling man to pieces, the falling man wouldcrush the victim to death.9

The falling man, Thomson argues, unintentionally violated a claim thevictim holds against him: he is under a duty to the victim not to killhim. In and of itself, this fact entails that the falling man is liable todefensive killing (as this is the only way to eliminate the lethal threathe poses).

As we shall see in Section II.C, there is an important differencebetween the way the right of self-defense is understood within Thom-son’s theory and the way it is understood within the traditional warconvention. To anticipate, the former theory does not allow for “moralequality”: in Innocent Threat, for example, the fact that the victim hasa right to kill the falling man entails that the falling man has no rightto kill the victim. In contrast, the conception of self-defense embeddedin the traditional war convention resists this entailment; it does allowfor moral equality. Before discussing this difference in more detail, Ishould like to point to the close relations between the Thomsonian andthe traditional visions of the right of self-defense.

For Walzer, the rules of conduct in war should “reflect as closelyas possible the same principles of justice and liability that govern con-duct outside of war.”10 He believes that the traditional war conventionsatisfies this constraint: soldiers in war are required to discriminate be-tween, on the one hand, men and women in uniform, who are—inwartime—combatants (whatever their specific job in the military), and,on the other, civilians or noncombatants. This requirement reflects thedeeper moral obligation to discriminate attackers from bystanders. Now,within a war, a soldier counts as posing a threat, by virtue of the pre-

9. Ibid., 287.10. This is Jeff McMahan’s formulation of his own position in his “Liability and

Collective Identity: A Response to Walzer,” Philosophia 34 (2006): 13–17, 13; Walzer makesclear that the war convention is “morally plausible” (see, e.g., Just and Unjust Wars, 133).


sumption that he will participate in the warfare.11 Soldiers’ status asthreat-posing individuals does not follow from the threatening acts theydo or from their culpability for the material threat the army to whichthey belong poses. Thus, moral innocence and lack of agency do notsecure immunity from defensive killing. Indeed, innocent, not quiteinnocent, and guilty enemy soldiers are all present in the battlefield.Notwithstanding these differences between them, soldiers are equallyliable to killing in wars. Thomson’s theory of self-defense provides anattractive explanation of the insensitivity the traditional convention man-ifests to such differences; individuals might become liable to defensivekilling, even if the threat they pose involves neither intentionality noragency.

B. Against the Moral Equality of Soldiers: McMahan’s First Argument

The traditional convention attributes to unjust combatants the right tokill just combatants in self-defense. This ruling, McMahan claims, hasno moral basis within Thomson’s framework (as it is sketched in theprevious subsection). After all, in this framework, liability to defensivekilling follows from violation of rights. And, normally, just combatantsviolate no right possessed by unjust combatants. McMahan’s argumentfor this claim is simple and effective: in the paradigmatic case, an unjustwar is initiated by a surprise attack against “the members of [a] standingarmy that is based in its home-territory. . . . [This army] is not mobilizedand has no plans for war.”12 The attacked soldiers do not pose a threatto others. Hence, the initial attack on them is a violation of their rightnot to be attacked. This is inconsistent with the traditional just wardoctrine, according to which, insofar as he follows the jus in bello rulesof fighting, a combatant has a right to participate in a war, howeverunjust it is.

In fact, McMahan contends that unjust combatants never acquirea right to kill just combatants in self-defense (unless the just combatantsviolate their own duties, e.g., they attack materially innocent civilians).Unjust combatants do not differ from the thief in “Robbery”:

Robbery: During a robbery, the guard of the bank had tried to geta gun in order to neutralize an armed thief. The guard’s preventiveactions were not disproportionate. Realizing that this was theguard’s intention, the thief shot the guard and killed him. Initially,the thief had no murderous intentions. His killing was purely de-fensive: had he not killed the guard, the guard would have killedhim.

11. This is one of main theses in Michael Walzer, “Terrorism: A Critique of Excuses,”in his Arguing about War, 51–67, and “Terrorism and Just War,” Philosophia 34 (2006): 1–12.

12. McMahan, “Ethics of Killing in War,” 697.


Needless to say, the thief had no right to kill the guard. This is as truein the Thomsonian framework as it is in every other (non-Hobbesian)framework; by taking advantage of his right to stop the robbery, theguard does not lose his right not to be attacked by the thief. This iswhy the thief’s killing counts as murder. “Most find it impossible tobelieve that, by unjustifiably attacking you and thereby making it jus-tifiable for you to engage in self-defense, your attacker can create theconditions in which it becomes permissible for him to attack you. Mostof us believe that, in these circumstances, your attacker has no rightnot to be attacked by you, that your attack would not wrong him in anyway, and that he therefore has no right of self-defense against yourjustified, defensive attack.”13 I shall refer to the argument in this passageas McMahan’s First Argument.

McMahan’s First Argument concludes with a “purist” war ethic.Purists contend that the killings committed by unjust combatants arewrongful homicides (which might be excused to some degree), whereas,if necessary and proportionate, the killings committed by just combat-ants are morally permissible or even morally justified.

C. Against Collectivized Morality: McMahan’s Second Argument

The collectivized morality of the war convention is claimed to be afurther Hobbesian aspect of the traditional just war doctrine. AsMcMahan reads him, Walzer views the common beliefs about individualmorality outside the context of war as largely irrelevant to the conductof war: “[Walzer] embraces the accepted view that ‘it is entirely legiti-mate to kill soldiers at random, as they come within range, so to speak,and it is legitimate to try to terrorize the ones who never come withinrange.’”14 McMahan then asks, “What is it about soldiers, once war hasbegun, that makes them ‘rightly subject to attack, all of them, all thetime’ (unless they have surrendered or been taken prisoner)?”15 “Ac-cording to Walzer’s view—and here he speaks for the dominant traditionwithin just war theory—it is because they are ‘materially engaged in thewar effort.’”16 The difficulty McMahan brings out is deeply disturbing:many soldiers are not engaged in fighting the war. Why are these soldiersliable to being killed? Walzer answers that they are all the members ofan army and “contribute to the achievement of its ends. . . . In timeof war they pose a unified threat.”17 This has to do with the fact that

13. Ibid., 699.14. Walzer, “Terrorism and Just War,” 1.15. Ibid.16. McMahan, “Liability and Collective Identity,” 13.17. Walzer, “Terrorism and Just War,” 1.


soldiers are collectively organized and disciplined, trained in the use ofweapons, and isolated on bases provided for them by the state.

McMahan understandably complains that even if all of these char-acteristics are correctly attributed to the individual soldier, none of thementail that, in time of war, he makes a material contribution to the threatthe army poses. A military organization has many goals: “Some . . . areentirely unrelated to the war. Some [soldiers] may even obstruct theprosecution of the war: for example, a military lawyer who, acting infidelity to her understanding of the military’s own guiding principles,seeks to revise the rules of engagement in a way that would hamper theeffectiveness of military action in this instance.” Hence, a Thomson-styletheory of self-defense does not support the convention in which totallyinnocent soldiers are liable to defensive killing. However, “some peoplewho are uncontroversially civilians do contribute, sometimes more sig-nificantly than most soldiers—for example, a scientist employed by auniversity whose research is financially supported by the Department ofDefense because among its many possible applications are certain im-provements in weapons technologies that can be used to great advantagein an ongoing war.”18 I shall refer to this argument as McMahan’s SecondArgument.

II. THE MORAL EQUALITY OF SOLDIERS AND A LOCKEANCONCEPTION OF SELF-DEFENSE

The case for the moral equality of soldiers I shall advance will be twostaged. In this section I show that the theory that best explains thepermission to kill unjust combatants is based on Lockean convictionsand that this theory of self-defense entails that in many typical casessoldiers of both sides have an equal right of self-defense. The very pos-sibility of such a theory counters the purist accusation that the warconvention embeds a Hobbesian vision of this right. In Section III, Ishow that “contractual considerations” (that do not follow from theethics of self-defense but are strongly suggested by my rights-based con-ception of it) imply that soldiers are morally equal in all other cases.

A. Robbery, Sleeping Soldiers, and Sleeping Untrained Soldiers

Robbery is a misleading analogy, and the purist critique of the traditionalconvention inspired by this analogy is therefore mistaken.19 In order toshow that, I shall describe two cases—“Sleeping Soldiers” and “SleepingUntrained Soldiers”—through which the problematic consequences ofpurism become more visible. Consider first,

18. McMahan, “Liability and Collective Identity,” 14.19. This section draws on my “Responsibility of Soldiers.”


Sleeping Soldiers: A war has been initiated by an unjust aggressioncarried out by an elite (and very small) unit of unjust combatants.The preparations for the military campaign were highly confiden-tial. Very few unjust combatants were aware of them. Now, imme-diately after the surprise attack had been launched, just combatantsresponded by attacking sleeping enemy soldiers. At the time thesleeping soldiers joined the army, it was permissible to do so. Ad-ditionally, they contributed nothing to the unjust attack, knew noth-ing about it in advance, and are not responsible for the aggressioninitiated by their country in any other way. Even so, presumably,they will participate in the war. So, their killing is preventive and,as such, a means for achieving a legitimate military goal.

Purists face a dilemma. According to the purist ethics of self-de-fense, liability is intrinsically related to agency. And, on the face of it,the sleeping soldiers did not do anything that makes them liable tokilling. Hence, purism seems to entail that the sleeping soldiers are notliable to killing and just combatants ought to avoid targeting them. Soit seems that in order to sustain their core conviction with respect tothe impermissibility of the killing of unjust combatants, purists have toadvance a radical reform of the rules of conduct to which just com-batants ought to subject themselves.

But suppose that the purist ethics of self-defense is more permissive;the fact that the sleeping soldiers will participate in the unjust war makesthem liable to defensive killing. Or, alternatively, these soldiers lost theirimmunity due to the fact that they would have been ready to join thefighting forces. Or perhaps the sleeping soldiers are responsible for theunjust threat they now pose because when they joined the army theyknew that they might participate in unjust wars.20 This move is quitecostly as well; it commits purism to deny the sleeping soldiers a rightof self-defense. For, the purists’ other core conviction is quite clear: likeall other cases, Sleeping Soldiers involves no moral equality; if the justcombatants’ preventive attack is permissible, the unjust combatants haveno right to eliminate the threat just combatants pose by killing them.That is, the sleeping soldiers are denied a right to self-defense, despitethe fact that their membership in the military forces is permissible, theyare not yet the agents of any threat, and they have had no chance toopt out from the unjust war. The dilemma is straightforward: purismdenies either that the preventive killing in Sleeping Soldiers is permis-sible or that the sleeping soldiers retain their right of self-defense. Bothclaims are counterintuitive.

20. A close analysis of Jeff McMahan’s writings suggests that this would be the lineof argument to which he would adhere. Compare his “Self-Defense and the Problem ofthe Innocent Attacker,” Ethics 104 (1994): 252–90, 260, to his “Ethics of Killing in War,”723.


In Sleeping Untrained Soldiers, purists confront a harder dilemma:

Sleeping Untrained Soldiers: Like Sleeping Soldiers, except that theunjust combatants have just been drafted and there is no way totrain them before the current war ends, they are attacked in orderto terrorize and deter other unjust combatants. Thanks to this(counter-) attack, the army—whose soldiers were demoralized bythe first strike—is back on its feet again. In addition, the attack isproportionate; there is no “cheaper way” to achieve the legitimateaims listed above.

Purists are more likely to bite the bullet in this case: given the totalinnocence of the targeted soldiers in Sleeping Untrained Soldiers, kill-ing them is as wrong as killing totally innocent civilians. But then, purismwould require obtaining specific knowledge about how threatening un-just combatants are, as a precondition for attacking them. Again, underthis reading of purism, the reforms it advances are far reaching, un-realistic, and excessively high minded.

Purists might come up with a conception of self-defense that some-how permits targeting the soldiers in Sleeping Untrained Soldiers. Pre-sumably, these soldiers would have been ready to participate in theunjust war—had they been trained and ordered to do so. Purists mightallude to this fact as the basis of the liability of these soldiers. True, thesleeping soldiers won’t be part of any further attack. But this, puristsmight argue, is accidental and hence morally irrelevant. So understood,the purist ethics of self-defense seems unreasonably permissive; it allowstargeting nonthreatening soldiers on the basis of the counterfactualthat had they been trained, they would have joined the fighting troops.Could counterfactuals of this type be a basis for the liability to defensivekilling? This seems unlikely: many civilians would have been ready tofight an unjust war, had they been trained and asked to do so. Worse,this permissive version of purism implies that the sleeping soldiers haveno right of self-defense in spite of their total innocence. The dilemmais again clear: purism denies either that the preventive killing in SleepingUntrained Soldiers is permissible or that the sleeping soldiers retaintheir right of self-defense. With respect to this case, both claims seemeven more counterintuitive.

I won’t discuss the purist approach to these cases any further. Fornow, I also put aside Sleeping Untrained Soldiers (a case I shall discussat length in Sec. III.D). The very dilemma purism faces in SleepingSoldiers conveys what is important for my purpose in this part. In at-tributing the right of self-defense to the sleeping soldiers, we seem tobe guided by Lockean convictions as to whether these soldiers havedone anything that could have caused them to lose this right. OurLockean intuition is clear: the fact that the sleeping unjust combatants


are not responsible or minimally culpable for the threat the army (towhich they belong) poses guarantees their right to self-defense. Theseconvictions have nothing to do with the Hobbesian view, according towhich one is allowed to do everything one takes to be necessary forone’s survival.

Note, however, that I do not deny another clear intuition: the sur-prise counterattack against the sleeping soldiers is justified; I merelyassume that, notwithstanding the justification of the counterattack, thesleeping soldiers retain a right to defend themselves. I shall put thiscommonsensical assumption in the language of rights: in Sleeping Sol-diers, soldiers are morally equal because both sides lost the right notto be attacked by each other but retain their right to defend themselvesfrom these attacks. In the following subsections I systemize these con-victions and argue for them; I develop a Lockean, rights-based theoryof self-defense that yields the above (appealing) description of the nor-mative reality in Sleeping Soldiers and in a wide range of other cases.I shall also argue that Sleeping Soldiers is not exceptional and that warsshould be analogized to this case rather than to Robbery. That is, theethics of self-defense that yields the correct result in Sleeping Soldiersentails that, within many circumstances that wars are likely to generate,just and unjust combatants are morally equal.

The argument unfolds in the following way. Section II.C sketchesa crucial modification of Judith Thomson’s original theory of the rightto self-defense. In Sections II.D–E, I shall use the modified conceptionin order to elucidate our moral intuitions in cases that involve moralequality. In Section II.F, I show that wars are likely to generate circum-stances in which, according to the rights-based ethics developed in Sec-tions II.C–E, soldiers are morally equal. I begin with constructing theconceptual framework that underlies the theory of self-defense that Ishall develop.

B. Preliminaries

The normative framework that best describes the distribution of rightsand duties in Sleeping Soldiers is a modification of Thomson’s concep-tion of self-defense.21 I shall call it SD. The moral facts to which SDappeals are expressed by sentences of four types:

1. X has a right to A (usually, Y in self-defense).A p kill2. X is justified in A-ing (usually, Y in self-defense).A-ing p killing3. The distribution of harm is un/just. The threat posed by X is un/

just, and

21. For a further defense of this conceptual framework, see my “Culpable Bystanders,Innocent Threats, and the Ethics of Self-Defense,” 615–21.


4. X ought, all things considered, (not) to A. X is morally obliged toA/avoid A-ing.

SD employs other moral notions, which are definable by the notions of(1) right, (2) justification, (3) just distribution, and (4) moral obligation.But, unfortunately, these basic notions are terribly vague. In this sub-section I clarify what I mean by saying that X has a right to A, is justifiedin A-ing, ought (not) to A, or poses an un/just threat. I do not pretendthat my use is the most prevalent one, nor do I argue that this is howthe concepts that appear in 1–4 should be used in all contexts. Whatfollows is mainly a series of meaning stipulations. The framework thatemerges paves the way for SD, which is, I argue, the theory that bestcaptures our intuitions with respect to the right of self-defense.

Consider then the proposition expressed by “X has a right to A.”Formally, the right to A is constructed as a moral privilege: X has a rightto kill in self-defense if for all Y, X violates no claim Y has against him.(Hohfeldian right claims and duties will be referred to as ‘claims’ and‘duties’, respectively.) I additionally assume substantive intrinsic rela-tions between a person’s legitimate interests and the rights this personpossesses. In particular, the right to self-defense is a right to be partial.A victim who exercises this right avoids the harm the attacker is aboutto inflict on him by shifting it to the attacker. Yet, rather than enforcingjustice in the distribution of harm, or bringing about an outcome thatinvolves less evil, the agent merely protects his own self-interest.

Thus, facts about rights of self-defense can be put in terms of “weakmoral permissibility”: A is (weakly) permissible if and only if, by A-ing,the agent violates no claim held against him. The weakness of this notionof permissibility is revealed by the fact that (as a matter of conceptualpossibility), even if by A-ing the agent would act permissibly, it is notimpermissible to prevent him from A-ing. Hence, conceptually, boththe victim and the attacker in Innocent Threat (say) might have a rightto kill each other in self-defense. Weak permissibility is quite prevalent:defeating your rival in chess is morally permissible. Clearly, it is alsopermissible for your rival to prevent you from defeating him. Anothercrucial feature of weak permissibility is that it is a middle-ground cat-egory; the fact that X’s A-ing is weakly permissible does not imply thatX’s A-ing is justified, yet it does imply that it is more than merely ex-cusable. And the fact that X’s A-ing is more than merely excusable hasimportant consequences. As a middle ground between excuse and jus-tification, permissibility generates moral immunity from interference bycertain people and institutions; for example, if A-ing is weakly permis-sible, an impartial institution has no moral power to criminalize it.

As for 2, I define the notion of justification very narrowly, in a waythat relates it to impartiality. I do so by first adopting the formal features


that characterize justification on most views. It is extendable to a thirdparty: if X is justified in killing Y in self-defense, anyone else is justifiedin doing so, in X’s defense. And, if X’s killing in self-defense is justified,it is unjustified to prevent X from doing it. More important, I shallstipulate that a person is prima facie justified in A-ing, if A-ing is rec-ommended by the principle of the lesser evil. That is, X is prima faciejustified in A-ing, if by doing so he brings about outcome 1 (O1) insteadof outcome 2 (O2), and O1 contains less evil than O2. Outcomes 2 and1 meet this requirement if the number of harmed people in O1 is smallerthan the number of people harmed in O2, or the magnitude of theharm people suffer in O1 is smaller than in O2, and all other thingsare equal. A final stipulation with respect to impartial justification: aperson is prima facie justified in bringing about O1 rather than O2, ifthe harm in O1 is more justly distributed than the harm in O2, and allother things are equal. As I use the notion, justification is sensitive tothese considerations only. Hence, I should admit that the notion as itis defined here is unusual.

I won’t provide a full account of justice in the distribution of harmor of the truth conditions of the schemata in 3. But some self-evidenttruths can be safely assumed. By definition, a threat is “unjust” if it isunjust that the victim of the threat—rather than the person who posesit—will be harmed because of it. Usually, shifting unjust harms enforcesjustice in the distribution of harm. For example, the thief in Robberyposes an unjust threat to the guard—and it would have been just toeliminate this threat by inflicting harm on the thief. But, there is acrucial complication here which Innocent Threat makes vivid: the fallingman poses an unjust threat to a victim. As the falling man is not theagent of the threat, he is not even minimally culpable for this threat.Therefore, the defensive threat created by shifting the harm to thefalling man is unjust as well. Put more generally, if an attacker is noteven minimally culpable for the unjust threat he poses, shifting theharm to him would be unjust as well.22

A final remark regarding justice in the distribution of harm andits relation to culpability: let us suppose that there are cases in whichan attacker ought, all things considered, to pose an unjust threat (weshall shortly see that there are such cases). If so, he is not minimally

22. As an Ethics’ editor remarked, one might readily agree that shifting harm ontosomeone who poses an innocent threat is no improvement in justice. We still might ask,however, why does that make it “unjust” to do that? After all, the shift might be neitherjust nor unjust. My answer: the harm inflicted on the attacker is unjust because it is amatter of brute bad luck. The inequality generated by this shift is, thus, unfair. This, ofcourse, calls for elaboration, which is beyond the scope of this article.


culpable for the threat he poses because he had no morally preferredalternatives. Hence, shifting the harm to him is unjust.

This final remark on distributive justice leads us to 4. My commentson the truth conditions of “X ought, all things considered, (not) to A”and on the notion of moral obligation are negative. Most important, Xmight have a right to A, even if X ought to waive it. Consider an enrichedversion of Innocent Threat where, in order to survive, the victim hasto kill five people who fall on him. Conceptually, it is possible that thevictim have a right to kill the five in self-defense, while being obligedto waive this right; he has a right to be partial, even if morality requiresimpartiality.

Furthermore, there is an important gap between the narrow con-cept of justification defined above and the inclusive notion of moralobligation characterized now. Suppose that by A-ing, X brings about anoutcome that involves less evil (such that in the narrow sense definedabove, he is justified in A-ing). Still, it might be the case that X oughtnot to A. This phenomenon manifests the complexity of the morallandscape: X is subject to morally valid contractual and special obliga-tions. In some circumstances, these obligations imply that X ought tobring about an outcome that involves more evil. For example, it is widelybelieved that firefighters ought to save those who pay for their service.They ought to do so, even in circumstances in which they can save alarger group of people to whom they bear no contractual duty. Similarly,it is widely believed that X ought to save his children from the fire, evenif he could save many others, who are unrelated to him.

The notion of moral obligation is inclusive; it is sensitive to rights(to be partial), to impartial justifications, to contractual and specialduties, and so on. As such, it commands a notion of strong moral per-missibility. “A-ing is morally permissible” means that it is not the casethat the agent ought, all things considered, not to A. Unless explicitlystated otherwise, in what follows ‘permissible’ means ‘weakly permissi-ble’.

C. SD and the Moral Equality in Innocent Threat

As noted, I call the rights-based theory I develop ‘SD’. It is a conjunctionof two propositions, SD-1 and SD-2. SD-1 follows from Thomson’s theoryof self-defense, while SD-2 does not and instead modifies it. Both com-bine to form the view endorsed here.

SD-1: Suppose that Y constitutes an unjust threat to X’s life andthat Y has no right to do so. Then, X has a right to kill Y in self-defense. That is, by killing Y, X violates no claim X holds againsthim (if the killing is necessary and proportionate).

According to Thomson, having a right to kill in self-defense entails


being justified in doing so. And vice versa: for Thomson, any instanceof injustice necessarily involves a right violation. She thus believes ofInnocent Threat that, from the fact that a victim has a right to kill thefalling man in self-defense, it follows that he is justified in doing so.And this implies, on Thomson’s notion of rights, that the falling manhas no right to eliminate the defensive threat imposed on him. Butthese results are implausible. It seems obvious that the falling man doeshave a right to eliminate the defensive threat posed by the potentialvictim.

Within the conceptual framework that underlies SD, the right thevictim has does not entail that his killing is justified. Moreover, the fallingman is not an agent of the threat he constitutes, so he is not culpablefor it; so shifting the harm to him does not implement justice in thedistribution of harm. The defensive threat posed to him by the victimis, therefore, unjust. In fact, then, the conceptual framework developedin the previous section strongly suggests that SD-2 is true. I take it tobe a crucial correction of Thomson’s theory:

SD-2: Suppose that at time , Y poses a threat to X’s life, aimingt � 1to eliminate an unjust threat posed by X at t. Then,

a) If X is not even minimally culpable for the threat he posed att, the defensive threat Y poses is unjust. Hence, subject to con-straints of proportionality and necessity, X has a right to elim-inate the defensive threat that Y poses, by killing Y.

b) If X has no right to pose the threat he actually poses and he isculpable for posing it, then the defensive threat posed to himis just. X, therefore, has no right to eliminate the defensive threatimposed on him by killing Y.

In the Thomsonian view I adopt, the falling man in Innocent Threatposes an unjust threat to the victim and has no right do so. By SD-1,the victim has a right to kill the falling man in self-defense (i.e., it isweakly permissible for the victim to do so). But, by SD-2, if the victimtakes advantage of the right he holds, the falling man has a right toeliminate the defensive threat imposed on him by killing the potentialvictim. (Hence, eliminating the defensive threat the victim poses is alsoweakly permissible.) The reason is simple: the falling man is not evenminimally culpable for the threat he poses, so the defensive threat im-posed on him is unjust. The victim, I suggest, has a right to preventone injustice by doing another. In this sense, the victim and the fallingman in Innocent Threat resemble chess players—they both have theright to defeat each other. In sum, Innocent Threat is a case of moralequality; both, the falling man and the victim, lost their right not to beattacked but retained their right to self-defense.


D. Further Motivating SD: Moral Equality in “Symmetrical Case”

Interestingly, Innocent Threat resembles a simpler and clearer case ofmoral equality.

Symmetrical Case: X and Y are driving two trains toward each otheron the same track. Following the instructions of a negligent in-spector, they started driving at the same time. A total-loss crashwhere both drivers are killed is bound to happen, unless one driverstops the other by using deadly force. They cannot avoid the crashby stopping the trains. For, upon realizing the danger, the driversare too close to each other.

The “optimal” result of Symmetrical Case is an outcome in which eitherX or Y is killed but not both, whereas “the suboptimal result” would bethe outcome in which both are killed. What is each driver permitted todo? One possibility is that X and Y should let the coin decide who willsurvive the tragedy. The procedure is fair, and if X and Y can com-municate, cooperate, and trust each other, randomization secures anoptimal result. Alas, in Symmetrical Case, X and Y cannot communicate.And, a noncooperative randomization increases the chance that bothdrivers will avoid action. Free competition seems a safer way to reachone of the optimal outcomes.

It might be thought that rather than optimal, randomizing is fairby virtue of the ex ante equality it secures. If X and Y let a coin decidewho will survive, their expected utilities are equalized. Does the ex anteequality make the ex post unequal distribution in one of the optimaloutcomes just? There is no reason to think so; it would be unfair to Xif he is killed just because of the brute bad luck that he was slower thanY. It would be as unfair to him if the coin decides against him. Indeed,there is no fair optimal outcome in Symmetrical Case—and the fairnessof the ex ante equal distribution of chances to survive won’t change thisfact.

I thus suggest that X and Y each have a right to kill the other inself-defense and believe that most of us would accept this suggestion.Both, X and Y, have a right to act in their own (conflicting) interests,despite the fact that neither’s interest is weightier than the other’s;hence, neither is (impartially) justified in doing so. It follows that noaction could enforce just distribution in Symmetrical Case. Let me re-emphasize that the killings both agents have a right to commit are notmerely excused. A killer is excused because he is not to be blamed orpunished for this right violation. But the excuse/justification dichotomyis too crude: having a right implies more than having a valid excuse but


less than having justification.23 Like Innocent Threat, Symmetrical Caseexemplifies a very complicated cluster of normative facts:

1. By imposing an unjust threat on Y while having no right to do so,X lost the claim he had against Y that Y won’t attack him in self-defense. Still,

2. X is not even minimally culpable for the threat he poses; hence,X retains the right to defend himself against Y’s defensive attack,even if it means killing Y.

Mutatis mutandis, the same is true of Y.

E. SD and the Moral Equality in Asymmetrical Cases

Might a typical war generate circumstances in which both sides lose theright not to be attacked but retain the right of self-defense? The purists’answer is negative. They probably believe that, as a matter of conceptualnecessity, cases in which both parties enjoy a right of self-defense are,if not strictly symmetrical (like Symmetrical Case), morally symmetrical(like Innocent Threat). I now want to show that—contrary to purism—moral equality in asymmetrical cases like Sleeping Soldiers is possible.

Consider the commonsense view regarding “Trolley.”

Trolley: A runaway trolley is careering down the mainline track. Ifit continues along this track, it will crash into the station, killinghundreds of people. It can be diverted onto a branchline track onwhich stands an innocent person, the victim. You stand on thebridge just above the branchline track. And you have access to theswitch that can divert the trolley. You do so. The trolley is now onthe branchline track, so you already saved the hundreds. It is toolate for the innocent bystander to run away. Yet, he can defendhimself by shooting you. For, as a result of the shooting, you willfall on the branchline track and stop the trolley before it hits him.In other words, it is too late for the victim to run away but not tokill you in self-defense.

Most people believe that you ought to divert it (in the inclusive sensedefined in Sec. II.B). In such a context, the principle of lesser evil seemsto them most weighty. I will grant this common view but assume—again,in accordance with commonsense morality—that the harm inflicted onthe bystander is unjust. It seems clear that the bystander retains his rightof self-defense.

SD has a simple explanation for this conviction. In diverting thetrolley you impose an unjust threat on the victim, and you had no right

23. Compare Russell Christopher, “Self-Defense and Defense of Others,” Philosophy& Public Affairs 27 (1998): 123–41, and “Self-Defense and Objectivity: A Reply to JudithJarvis Thomson,” Buffalo Criminal Law Review 1 (1998): 537ff.


(against him) to do so. By SD-1, the victim acquires a right to kill youin self-defense. However, you ought to divert the trolley; that is, divertingthe trolley is the morally preferable alternative you face. So, in an im-portant sense, you are not even minimally culpable for the unjust threatyou impose on the bystander; you had no morally preferred alternative.Under this interpretation of “minimal culpability,” SD-2 entails that (sub-ject to the constraints of proportionality and necessity) you have a rightto eliminate the threat posed by the bystander, by killing him.

Trolley involves moral asymmetry: in the narrow sense articulatedby the principle of lesser evil, you justifiably impose an unjust threat onthe victim, while the victim has no parallel justification for imposing anunjust threat on you. This moral asymmetry notwithstanding, both ofyou have an equal right of self-defense: both of you lost your right notto be attacked but retained your right to defend yourselves from theseattacks.

Consider another famous case, in which the moral asymmetry iseven sharper:

Tactical Bomber : A pilot fighting a just war has been ordered to bomba munitions factory located on the border of the enemy country. Heknows that if he bombs the factory, the explosion will kill the groupof innocent civilians just across the border in a neutral country. Butthis would be a side effect, and the harm to the civilians would beproportionate to the good of the destruction of the factory—in thelong run, the bombing would save the lives of thousands.

In the narrow sense of justification used here, the bomber is justifiedin posing an unjust threat to the innocent civilians. If he is successful,he would bring about an outcome that contains less evil. Moreover, Ishall grant the received view that, given the numbers of the lives savedby the bombing, the role of the bomber in the army, and the bomber’sintention and efforts to save lives rather than to kill innocent neutrals,he ought, all things considered, to undertake the mission. Therefore,if the civilians kill the bomber and cause the failure of the bomber’smission, they bring about an outcome that contains more evil.

Still, as in Trolley, there is no real intuitive doubt that, despite beingjustified in undertaking the mission, the bomber had no right (againstthe civilians) to threaten their life and that the threat he poses is unjust.It follows from SD-1 that the civilians have a right of self-defense; thatis, in killing the bomber in self-defense, they violate no claim he holdsagainst them. They possess this right, even if it is further assumed thatthey are perfectly aware of the facts that justify the bombing of thefactory. In other words, they hold a right to attack the bomber, despitethe fact that in doing so they bring about an outcome that involvesmore evil. That the bomber retains a right of self-defense seems evident


as well. That is, if the civilians attack him, he has a right to directly killthe civilians—even if, as a result, he cannot complete the mission. Inlight of my above interpretation of minimal culpability, SD-2 is a readyexplanation for this: as the bombing of the factory is something thepilot ought to do, it is the best alternative he had at the time. Hence,he is not minimally culpable for the unjust threat that he poses.

No doubt, the thief in Robbery is very different; he has no rightof self-defense while being engaged in the armed robbery. Why? Whyis he so different from the unjustified civilians in Tactical Bomber? Howdid he lose his right of self-defense? The plausible answer follows fromSD-2: unlike the agents in Innocent Threat, Trolley, and TacticalBomber, the thief is culpable for the unjust threat he poses.

F. SD and the Moral Equality of Soldiers

So understood, SD suggestively articulates and explains the moral equal-ity in Sleeping Soldiers. Consider a core Thomsonian conviction: a per-son might lose his title to life, even if he is not the agent of the lethalthreat he constitutes. In accordance with that conviction, I suggest thatthe sleeping soldiers in Sleeping Soldiers constitute a threat by virtueof the true presumption that they will participate in the war. If so, SD-1 entails that just combatants have a right to kill them in self-defense.It is, however, the corrupt political leader who pushed unjust combatantsinto the battlefield. They are not the agents of the threat they pose, sothey are not culpable for it. Therefore, it follows from SD-2 that theyretain a right to defend themselves. In other words, Sleeping Soldiersshare with Tactical Bomber a crucial feature: they are both morallyasymmetrical cases that involve moral equality. Like the bomber (in thelatter case), just combatants (in the former case) fight for an (impar-tially) justified aim; they try to bring about an outcome which containsless evil and in which harm is distributed more justly. Still, like thebomber, the threat they pose is unjust, hence, the resemblance betweenthe sleeping soldiers and the civilians in these cases; both have a rightto eliminate the unjust (defensive) threat posed to them.

I shall argue now that typical wars generate circumstances thatresemble Sleeping Soldiers. The basic reason was offered in passages inJust and Unjust Wars that McMahan quotes but, to my mind, misinter-prets. Walzer concedes that unjust war is a crime; it is, however, thecrime of the political leader. As for the soldiers, “their war is not theircrime”; “for the war itself . . . soldiers are not responsible.”24 McMahanagrees but claims that, at best, these facts form an excuse for the criminalviolence unjust combatants exercise. I read Walzer’s remarks differently.Unjust combatants’ lack of minimal culpability for the war they fight

24. Walzer, Just and Unjust Wars, 37–38.


explains why their killings within the war are defensive and why theyhave a right (of self-defense) to commit them.

How can an act of killing within a war of aggression be defensive?Observe, first, that before a war starts, there is a time at which it isepistemically determined that it is about to start. At that time, mostcombatants cannot know about the future war. Still, at that point, it ispresumably true of just and unjust combatants alike that they will fighteach other. So the analysis of Sleeping Soldiers I have just offered impliesthat they threaten each other at that time. Hence, a preemptive attack,whose aim is eliminating the unjust threat posed by unjust combatants,is permissible, even if the attack targets combatants that haven’t yetdone anything wrong. Soldiers are morally equal (partly) because of afurther complex fact: at the time in which it is determined that a waris about to start, it is also true of just combatants that they threatenunjust combatants. And, since at least some unjust combatants are noteven minimally culpable for the initial threat they pose, the defensivethreat posed to them is also unjust. By SD-2, these unjust combatantshave a right to eliminate this threat. They retain, in other words, a rightof self-defense, despite the fact that they lost the right not to be attacked.Hence, the moral equality of soldiers in such asymmetrical circum-stances: both sides lost their right not to be attacked by each other butretained their right of self-defense.

I shall develop these claims by looking closely at McMahan’s storyabout the surprise attack launched against a “standing army that is basedin its home-territory. . . . [The army] is not mobilized and has no plansfor war.”25 From SD’s standpoint, McMahan’s story is radically under-specified. There are circumstances that satisfy his description of whichSD says that the unjust combatants involved have a right (of self-defense)to participate in the surprise attack.

Suppose, for example, that at the time combatants joined the armyit was morally permissible to do so and that, unexpectedly, they findthemselves in the unfortunate situation in which it is (epistemically)determined that an unjust war will take place. No individual soldier orindividual act of refusal can prevent the unjust war. Sooner or later,unjust combatants will be counterattacked by just combatants, and noth-ing that the unjust combatants can do might stop this ( justified) attack.Hence, if unjust combatants are not minimally culpable for the threatthey initially posed, they have a right to eliminate the threat posed tothem. Furthermore, if auxiliary conditions (which I shortly spell out)are met, they have a right do so by participating in the surprise attack.I shall argue, in other words, that typical wars generate complex asym-metrical cases, which nevertheless involve moral equality.

25. McMahan, “Ethics of Killing in War,” 697.


Before presenting the argument in more detail, let me try to preventa basic misunderstanding. SD does not imply the repugnant conclusionthat a murderer (the unjust combatant) has a right to kill his victim(the just combatant) on the grounds that the victim’s acting in self-defense would pose a threat to the murderer. Remember that SD-2 saysthat an attacker has a right to eliminate the defensive threat the victimposes only if he is not even minimally culpable for the threat he initiallyposes. Thus, in Innocent Threat, the attacker (the falling man) becamea threat unintentionally; in Trolley and Tactical Bomber, the attackerposes a threat only because he did not have any morally preferredalternative to doing so. The example I shall build up combines thesetwo types of cases. Rather than the unjust combatants themselves, it isthe corrupt leader who created the initial threat they pose. Hence, unjustcombatants have a right to eliminate the defensive threat posed to themby just combatants. And, in the circumstances I shall describe, they havea right to eliminate this initial threat by participating in a surprise attack.

Consider, then, a surprise attack, which satisfies the following con-ditions:

1. The surprise attack is unjust.2. Still (despite appearances to the contrary), from the standpoint of

an individual unjust combatant, the surprise attack is defensive:the standing army poses a threat to him.

3. The threat that the standing army poses is unjust.4. The unjust combatant is not minimally culpable for the unjust

threat he poses by mounting a surprise attack against the standingarmy because he has no morally preferable alternative to doingso.

There are circumstances that satisfy 2–4, which I shall exemplify by ashort story. You are the hero of this story: you joined the army becauseyou believed that you would contribute to the security of your country.At the time, your belief was perfectly reasonable. While serving, youwere required to participate in a surprise attack, which you knew wouldbe the first event of an unjust war. Due to the state’s intensive propa-ganda, your fellows disagree with your judgment as to the morality ofthe attack. (These soldiers are indirectly responsible for the future war;the corrupt political leadership could not have counted on them if theywere more sensitive and critical.) You, however, are not one of the mass.You know that your leader is corrupt and that the war he initiates isunjust.

Alas, your objection to the war does not change a simple fact: apreemptive attack by the enemy soldiers is now permissible. For, it isnow (epistemically) determined that a future war is about to be waged;you and your fellows became a threat by virtue of the fact that you will


presumably attack in the near future. Now, you—as opposed to thosewho believe the war they are about to fight is just—are not really dif-ferent from the unjust combatants in Sleeping Soldiers. You became athreat by doing nothing. The same is true of the just combatants: thewar is inevitable; nothing they are able to do would prevent it, so pre-sumably, just combatants will eventually attack. And, 3 is true: the initialthreat posed to you by just combatants is unjust. We shall assume thatthe surprise attack preemptively eliminates this threat.

There is, though, a crucial difference between the cases I am tryingto analogize: unlike the unjust combatants in Sleeping Soldiers, you caneliminate the threat posed to you by fleeing from the battlefield or byrefusing to participate in the war. Hence, it might be thought that, asan exercise of the right of self-defense, the surprise attack is neithernecessary nor proportionate. You can eliminate the initial defensivethreat by refusing to participate in the surprise attack. It might bethought, in other words, that 4 cannot be true of any instance ofMcMahan’s story.

Admittedly, in some versions of McMahan’s story, refusal and sur-render are possible and morally preferable. Of such cases, 4 is false. Yet,these circumstances are quite circumscribed; usually, your personal sur-render is not the morally preferable alternative you face. For, the warunjustly put innocent people at risk, and these people count on yourdefense. As coordinated refusal cannot be arranged, some of yourfriends in the army are as innocent as you are. Further, the just sideimposes a dire risk on innocent civilians, even if, unlike the Allies inWorld War II, it fights in strict accordance with the war convention. Theconvention outlaws targeting innocent civilians. But, in some cases, itdoes permit killing civilians as a foreseeable side effect of bombingmilitary bases or munitions factories. We can assume that, thanks toyour participation in the war (in general) and in the surprise attack (inparticular), the risk to these people diminishes. Given your contractualand special duties to these innocent civilians, refusal is not the morallypreferable alternative you have.

In this context, it is especially important to distinguish between thenarrow notion of justification and the inclusive notion of moral obli-gation that were defined in the preliminaries (Sec. II.B). Your partici-pation in the surprise attack does not bring about an outcome thatinvolves less evil or in which the harm is distributed more justly. In thenarrow sense defined in the preliminaries, your participation in thesurprise attack is unjustified. Still, in the situation in which you aretrapped, valid impartial considerations are not more important thanyour contractual and special duties. Hence, it is not the case that youought not to participate in the surprise attack. Therefore, you are not


culpable for the threat you pose by doing so. Indeed, these circum-stances seem very common in wars.

McMahan anticipates and rejects a similar rejoinder to his critiqueof the moral equality of soldiers. “It might be argued,” he says, “thatthe unmobilized military personnel have combatant status because theydo pose a threat by virtue of being prepared to use force if their countryis attacked.”26 He responds as follows, “If the fact that they would fightif attacked is sufficient to make them count as threats to others, thenanyone, military or nonmilitary, who is prepared to fight if his or hercountry is attacked, must count as a threat to others and, therefore, acombatant.”27 Note, however, that the SD-based argument is better thanthe one McMahan anticipates (and rejects). The soldiers of the standingarmy do not pose a threat merely by virtue of the fact that they areready to fight if attacked. Rather, they pose a threat by virtue of the factthat they will attack. Hence, rather than merely conditional combatants,the members of the standing army are future combatants.

Unfortunately, the basic idea from which the SD-based argumenttakes off—the idea that X has a right to kill Y by virtue of the fact thatY will kill him in the future—is problematic. By its logic—the objectorwould claim—killing babies on the grounds that they will grow up tobe soldiers would be permissible. But this objection misses an essentialcharacteristic of our ethical reasoning: as moral agents, we reason andact on the assumption that the occurrence of some future events ispresently undetermined. To see this, suppose that a person voluntarilyengages in a permissible but risk-imposing activity, such as driving a car,and suppose, further, that contrary to reasonable expectations andthrough no fault on his part, he became a lethal threat. Saying of thedriver that he ought to have avoided the driving altogether because ofits tragic consequences sounds bizarre. It sounds as weird to say thatthe fact that he could not have known the tragic consequences of hisdriving is merely a valid excuse. In the epistemic sense, when he startedthe driving, the consequences were not in existence—there was nothingto know. It turns out, I suggest, that epistemic indeterminacy mattersfor morality. From a moral standpoint, the threat you and your enemiespose to each other exists only at the time in which it was epistemicallydetermined that war is about to take place.28

Let me conclude by describing the normative reality created by thecircumstances that satisfy 1–4 in terms of moral permissibility. In suchcircumstances, the killings committed by unjust but inculpable com-

26. Ibid.27. Ibid.28. Compare ibid., 724, and my discussion of the risk imposing driver case in “Re-

sponsibility of Soldiers,” 564.


batants are permissible in the two different senses that were defined inthe preliminaries. First, the killings are weakly permissible because sol-diers in such circumstances have a right (liberty) to kill each other inself-defense. Yet, second, in participating in the war, the soldiers weimagine contribute to the defense of the innocent people that counton their protection; they fulfill a contractual duty they undertook.Hence, given the fact that they hold a right of self-defense, it is not thecase that they ought not to participate in the war—even if they are ableto eliminate the unjust threat imposed on them by fleeing or refusing.

I should warn against an overstated misreading of the theory ad-vanced here: it does not imply that in circumstances described by 1–4,moral but unjust combatants ought to participate in the unjust war (inthe inclusive sense defined in Sec. II.B). There is another possibility. Byfighting the war, unjust combatants further an unjust outcome, so theyface two mutually exclusive courses of action—participating in the waror fleeing from it (say)—where no alternative is morally preferable tothe other.

G. Summary

As a Lockean theory, SD presents a radically constrained conception ofself-defense. And it is by virtue of its Lockean underpinning that it allowsfor cases in which an attacker loses his right not to be killed whileretaining his right to eliminate a lethal defensive threat the victim poses.Furthermore, we saw that typical wars, in which only one of the partiesis just, are likely to generate cases that involve moral equality. The puristaccusation—which reads the moral equality of soldiers as an expressionof the unconstrained Hobbesian conception of self-defense—was provedto be baseless. The traditional just war doctrine should be understoodin light of a perfectly sensible Lockean conception of self-defense.

Note, however, that, as a defense of the traditional war convention,the argument is not yet complete. The SD-based argument shows thatthe purist critique does not establish a complete inequality between justand unjust combatants. But, by itself, it does not establish the full doc-trine of the moral equality of combatants. For, first, the permissibilityof the killing of the nonthreatening combatants in, for example, Sleep-ing Untrained Soldiers is not yet explained. After all, SD does not permitkilling nonthreatening soldiers. Second, for traditionalists, unjust com-batants that do not satisfy 4—and thus are minimally culpable for theunjust threat they pose—have a right to kill in self-defense. Indeed, theconvention takes the culpability of unjust combatants for the threat theypose to be irrelevant. Isn’t the indifference of the traditional doctrineto facts about innocence and responsibility an expression of a Hobbesianvision of self-defense? In the next section I shall give a negative answerto this question.


III. BEYOND SELF-DEFENSE: THE WAR CONVENTION AS ATACITLY ACCEPTED SOCIAL NORM

The basic idea is simple: rights can be lost if their holders freely subjectthemselves to a procedurally fair and mutually beneficial norm (con-vention) that commands it.29 Just combatants lose their right not to beattacked by unjust combatants because they subject themselves to a rulethat disregards the distinction between just and unjust combatants. Theydo so because they know that it is in their interest that this rule wouldbe commonly followed. In other words, they lose their title to life (inwar) by virtue of their (very indirect) consent, just like boxers who, byagreeing on certain fair rules, lose their right not to be attacked by eachother. Indeed, a boxer who crippled another boxer in order to win acontest violated no duty, even if what he did was unjustified. War is evena less friendly activity than boxing, but the moral basis of fighting andboxing is very similar.

As noted, the contractual argument developed in this section ex-plains the moral equality of soldiers in cases that cannot be explainedby SD, like Sleeping Untrained Soldiers. But it also explains the equalityof soldiers in cases that SD already covered. Notwithstanding this, ratherthan obviating the SD-based argument, the contractual argument heavilydepends on it. For, absent the support of SD, the convention would bemorally invalid: if self-defense in war was totally asymmetrical, the “con-tract” between soldiers would have been invalid. Combatants are treatedby the convention as if they are not minimally culpable for the threatthey pose. In effect, the convention applies SD-2 to all soldiers, irre-spective of the cause for which they fight. Furthermore, the nearly uni-versal acceptance of the in bello code frees soldiers from the duty toobtain knowledge that could have made them culpable for the threatthey pose. And, in light of this exemption, SD says that they are morallyequal.

I shall start by explaining the mechanism that causes persons tolose moral rights as a result of subjecting themselves to commonly fol-

29. The debt of the argument in this section to David Lewis’s analysis of conventionin his Convention (Cambridge, MA: Harvard University Press, 1969) is deep. In particular,following Lewis I assume intimate relations between conventions, rules, and norms. I amalso indebted to George I. Mavrodes, “Convention and the Morality of War,” in InternationalEthics, ed. C. Beitz, Marshall Cohen, Thomas Scanlon, and A. John Simmons (Princeton,NJ: Princeton University Press, 1985), 75–89, and to R. B. Brandt, “Utilitarianism and theRules of War,” Philosophy & Public Affairs 2 (1972): 145–65. A version of the boxing matchmodel, which I endorse, was developed by Thomas Hurka, “Liability and Just Cause,” Ethicsand International Affairs 21 (2007): 199–218. Jeff McMahan envisages this model in his“On the Moral Equality of Combatants,” Journal of Political Philosophy 14 (2006): 377–93,381–84. My version of this model was developed independently of Hurka’s and ofMcMahan’s; it also significantly differs from theirs.


lowed rules. I shall delineate the conditions these rules meet if theyhave such an effect (Sec. III.A) and then show that the in bello rulesform a convention of this sort (Secs. III.B–C).

A. On Losing Rights Involuntarily by Virtue of Tacitly AcceptingSocial Norms

Assume, with Nozick,30 that property rights are moral and natural andthat, in the standard case, a person relinquishes ownership of an articleby freely giving up his claim to it—he sells it, gives it as a present, andso on. Suppose, however, that John unknowingly lost an article he owns.The article has no identifying signs; John won’t be able to demonstratethat he owns it. Or suppose John was coerced to drop the article some-where and leave it unattended. Most legal systems would negate John’sownership, allowing whoever finds the article to appropriate it. Indeed,John relinquished ownership of the article before he gave up his claimon it. Nozick’s theory of natural property rights might embrace theabove legislation: by losing the article, the owner lost his moral standingin relation to it.

In explaining the change in the moral status of the article we shouldappeal to an important social fact: our behavior suggests that we subjectourselves to the legal rules that govern exchange of property rights. Wedo so because we believe (a) that the rules are fair, (b) that it is mutuallybeneficial that these rules would be commonly followed, and (c) that,since others know that it is mutually beneficial that these rules wouldbe commonly followed, these rules are, in fact, commonly followed. Thisis a set of true and justifiable beliefs; it constitutes a piece of commonknowledge. The legal treatment of lost articles exemplifies both theefficiency and fairness of rules that govern the exchange of propertyrights. First, the prohibition to use the article would be an utter waste(efficiency). Second—given the fact that many people will deceptivelyallege the article to be theirs—the fairest way to proceed is to allowwhoever finds the lost article first to acquire it. I suggest, in short, thatin losing the article, the owner lost his moral claim on it by virtue oftacitly accepting the legal rule that allow others to appropriate it. Therelation between morality and legality is subtle: the fair legal norm istacitly accepted, and this is why it is has a moral impact.

This is not a fully consent-based libertarian story: the rules we acceptand endorse are elements in an indivisible system—and most of us haveno detailed acquaintance with them. The conventions to which we freelysubject ourselves are known to experts (lawyers, legislators, etc.)—whereaswe know them by description only, as the legal norms with whose detailslawyers are acquainted. Moreover, as Lewis makes clear—the knowledge

30. Robert Nozick, Anarchy, State, and Utopia (New York: Basic, 1974).


of conventions might be very “poor”; they are only tacitly known tomembers in the relevant society; in many circumstances, these individ-uals won’t be able to articulate them.31 In a clear sense, therefore, Johnlost his claim on the article involuntarily in one version of the abovetale and under coercion in the other.

Now, consider a modification of Trolley. A runaway trolley is ca-reering down the mainline track. If it continues along this track, it willrun over five people who are walking on the track. The five disobeyedan explicit instruction: there was a sign that warns, “Danger, Do NotEnter.”32 It seems that by deliberately disobeying the instruction, thefive lose the right not to be struck by the trolley, if the driver can savethem only by killing an innocent bystander. They also lost their rightto kill the driver in self-defense. Maybe we can go further than this.What if the five got to the track by mistake? And what if they werecoerced to be there? Is it plausible that they lost their right of self-defense mistakenly or under coercion? In light of the above discussion,a positive answer is quite understandable.

B. The Right of Culpable Unjust Combatants to Kill in War

The traditional war convention, in general, and the equality of all sol-diers, in particular, are—I now wish to suggest—norms that soldiersaccept and endorse. The acceptance of the convention by just combat-ants frees their enemies from the duty not to attack them within a warof aggression.

Practically, soldiers are equal if—whatever the cause of the war theyfight—they are immunized from any criminal prosecution and possessthe same set of rights and duties. In practice, they are legally requiredto respect the jus in bello rules only. Soldiers accept these rules because(a) they believe the rules to be fair, (b) they believe that it is mutuallybeneficial that these rules would be commonly followed, and (c) theybelieve that, since soldiers and statesmen know that it is mutually ben-eficial that these rules would be commonly followed, the rules are com-monly followed. All these beliefs seem to be justifiable and true. Byaccepting the traditional war convention on the basis of these beliefs,soldiers free all other soldiers from the duty not to attack them in warsand acquire, in return, a right to participate in wars, whatever theircause.

The justification for the belief expressed in b is simple: soldiers andcivilians live in states, which form an anarchical society.33 They would

31. Lewis, Convention, 63.32. Thomson, Realm of Rights, 178.33. The phrase is Hedley Bull’s. See his The Anarchical Society: A Study of Order in World

Politics (Oxford: Oxford University Press, 1977).


be vulnerable to large-scale unjust threats unless their right to life isprotected by a powerful political entity. Since these threats are a per-manent aspect of the reality in the anarchical society of states, efficientand well-ordered armies are necessary for protecting people’s right tolife. Yet, a well-ordered military force is efficient only if the great majorityof its members are obedient. Indeed, in an efficient army, the loyaltyand reliability of the soldiers is unquestionable. Ultimately, this meansthat soldiers in a well-functioning army give up the right to autonomyand undertake a duty to obey any legal order, however poorly informedthey are with respect to the causes of the order. Yet, if soldiers wereliable to legal prosecution for participating in an unjust war, many ofthem would not immediately follow the orders delivered by the polit-ical leadership. Since the existence of efficient armies minimizes theextent to which soldiers and civilians are vulnerable to unjust threats,the outcome in which unjust combatants are immunized from all legalresponsibility for the killing they commit in wars is (ex ante) mutuallybeneficial.

Of course, soldiers have not relinquished all autonomy. Quite tothe contrary. They are entitled to disobey orders that manifestly violatethe rules of conduct in war; they, in fact, ought to do so. My contractualargument implies an obvious distinction: under a regime that obligessoldiers to ascertain that the cause with which they fight is just, armieswon’t be able to efficiently fight a just war, to form a credible deterrentthreat, or to wage a just preemptive attack. Under such a regime, armieswon’t be able to provide efficient national defense. In contrast, a regimethat obliges soldiers to follow only in bello legal orders does not sufferfrom these deficiencies. The fact that the in bello rules are commonlyfollowed does not obstruct armies from providing self-defense.

Soldiers, I suggest, tacitly believe in the above true story—hence,b is true; c is true because soldiers can easily see that (broadly speaking)other soldiers respect the legal immunity of soldiers and expect theirown states to do so as well. Finally, a seems to express a true belief, aswell: our interest in self-defense is legitimate. The rules that constitutethe war convention protect this legitimate interest.

In accepting the legal rule that disregards the difference betweenjust and unjust combatants, soldiers give up their moral right claim notto be attacked in wars and in return acquire the moral right privilegeto follow orders of their political leaders. So when a war starts, the moralstatus of soldiers is changed by virtue of their prior acceptance of therules that constitute the war convention. Did just combatants consentto be unjustly attacked? In a sense, yes, but this does not seem to bethe best description of the normative reality wars generate. Soldiers losetheir title to life in the way that John lost his claim on the lost article;they accepted ex ante mutually beneficial rules, and they are now harmed


by people who act in accordance with these rules. Thus, there is a sensein which they lose their right to life involuntarily, as a result of thedecision of a corrupt political leader. They did not do anything thatcaused them to lose it at this particular time. Further, some of themmight be forced to join the army; these soldiers lost their title to life asa result of coercion.34

It might be thought that the egalitarian convention does not reallybenefit all implicated parties. In particular, it does not benefit membersof military forces who protect a state that is very unlikely to engage inunjust aggressive wars. Such soldiers—the objection goes—would ben-efit from a different rule which allows blind obedience only to non-aggressive governments. I disagree. First, I take it that in accepting theconvention, soldiers of a state S consider the interests of soldiers thatwill serve in S’s military forces as if they are their own interests. Second,I assume that it cannot be known now how aggressive a state is likelyto be in the remote future. Hence, in assessing how beneficial a con-vention is to members in S’s military forces, S’s presently apparent peace-ful dispositions are irrelevant. The equalizing convention is justifiablybelieved to be mutually beneficial, even by soldiers of a presently non-aggressive state.

It might be further argued that were soldiers accountable for theviolence they exercise—were they required to justify the war in whichthey participate in a national or an international court—the number ofunjust threats in the anarchical society of states would be minimized.If this were true, the contractual argument for the traditional conventionwould have been considerably weakened. Yet, it would not be completelyundermined. Thus, suppose that, within the optimal institutional ar-rangement, soldiers would be legally unequal. Even then the outcomeproduced by the current egalitarian legislation might be a “suboptimalcooperative equilibrium”: although a better outcome might be imag-ined, no one would be better off had he alone followed another rule.Currently, the general acceptance of the egalitarian convention is, inthis weak sense, mutually beneficial.

It would be useful to restate the conclusion of this argument interms of (first) moral permissibility, (second) justification, and (third)moral obligation. The common acceptance of the war convention freescombatants from making sure that the cause of their war is just. There-fore, the killing they commit within the war is morally permissible inthe weak sense defined in the preliminaries (Sec. II.B). The weak per-missibility of their killings is important; it distinguishes their acts from

34. In my “Elucidating the Supreme Emergency Exemption” (unpublished manu-script), I describe circumstances in which soldiers have no right to undertake the duty ofobedience.


the excused murderer in Robbery; legal institutions have, therefore, nomoral power to criminalize these acts. Second, like the unjust combat-ants described in Section II.F, unjust combatants whose rights are cre-ated by the universal acceptance of the convention do not bring abouta just outcome or an outcome that contains less evil. Hence, put interms of impartial justification, they are not justified in realizing theirright. Yet, there is a crucial difference between the unjust combatantsdiscussed here and those described in Section II.F. For, it is definitelypossible that the former ought not to take advantage of the moral free-dom they gained by the common acceptance of the war convention.Soldiers might have a right to participate in an unjust war that theyought to waive.

C. Two Objections

Two objections to this analysis merit close attention.35 Both are triggeredby one of the limitations of the argument. First, the argument showsthat unjust combatants have a right to defeat the enemy by killing justcombatants. Yet, unjust war is likely to cause harm to civilians and neu-trals—and soldiers on one side have no moral power to confer a rightto inflict harm on these people. Hence, the objection goes, unjust com-batants are under the duty not to cause collateral damage and therebynot to participate in an unjust war. Second, when a war is unjust byvirtue of its goal, just combatants cannot release unjust combatants fromthe general obligation not to do injustice.

Both objections convey deep skepticism as to the validity of thedistinction between jus ad bellum and jus in bello.36 I cannot defend thedistinction here, and critics are clearly right that it is problematic andvague. The following preliminary remarks presuppose that the distinc-tion is valid, but they also might be a kind of first step toward an ac-ceptable reformulation of it.

Consider the second objection first. I already stressed that in somecircumstances, soldiers ought to waive their right to participate in anunjust war. Hence, in effect, the second objection leaves the contractualargument intact. Interestingly, however, unjust combatants might ex-ercise their right to contribute to the defeat of the enemy, while fulfillingthe obligation to inhibit the achievements of the unjust goals of thewar. Here is a simple example. You are a soldier in a colonialist army.The people under occupation justly fight for their national indepen-dence. Currently, however, their just war puts at sharp risk the national

35. Both are offered by McMahan, “On the Moral Equality of Combatants,” 382;McMahan takes them to be devastating.

36. In Yitzhak Benbaji, “The War Convention and the Moral Division of Labor,”Philosophical Quarterly (forthcoming), I defend the in bello/ad bellum distinction in detail.


security of your country. A defeat would signal political and militaryweakness: you reasonably predict that other hostile countries will be-come more aggressive and threatening because of these signals. So adefeat of the enemy is in your legitimate interest. Yet, a victory by yourside is likely to cause a continuation of the unjust occupation, which isexactly the goal of your corrupt leader. While holding a right to con-tribute to the defeat of the enemy, you ought to struggle for the endingof the occupation; the latter moral obligation does not imply that youought to waive the right of self-defense you possess. You can fulfill it inother ways.

Just combatants might confront similar complexities. In their case,the goal of defeating the enemy is (impartially) justified, rather thanmerely permissible. Still, their political leaders might have unjust goalsthat a victory would promote. Probably, they are obliged to block theachievement of these goals. Yet, almost needless to say, that obligationdoes not wipe out the just combatants’ right to contribute to the defeatof their enemy.

Let us now turn to the first objection. Unjust combatants are boundto harm civilians and neutrals; according to the objector, this entailsthat they have no right to participate in the unjust war. But this simplydoes not follow. The contractual argument shows that unjust combatantshold a right to fight in order to defeat enemy military forces. This is apermissible but unjustified goal. Now, a further question should beasked: What are the permissible means for achieving the permissiblegoal of defeating the enemy? My defense of the traditional war con-vention leaves this question open. Indeed, the stringency of the pro-portionality constraint might be sensitive to the justice of the ultimategoal of the war. Perhaps, then, a moral but unjust combatant would beready to contribute to the defeat of the enemy only if no civilians andno neutrals are harmed by the violence he exercises. This possibility isperfectly consistent with the contractual argument, whose modest con-clusion is, to repeat, that defeating the enemy is a permissible goal (forsoldiers); it leaves the issue of the permissible means unsettled.

D. The Liability of Nonthreatening Combatants to Killing in War

How does the collectivized morality of the traditional convention surviveMcMahan’s Second Argument? Why, in particular, are the nonthreat-ening soldiers in Sleeping Untrained Soldiers liable to killing?

The answer I shall advance relies on two observations. The first isnormative: in war, political leaders control the military forces and assuch pose a material threat to their enemy. Most theories of self-defensewould imply that they are liable to defensive killing by virtue of thisfact. In other words, the war convention immunizes engaged civiliansdespite their (preconventional) liability. The second observation is fac-


tual: the leaders who are directly responsible for the war could not bedealt with in a precise way like assassination or bombing of (selected)government buildings. For, the responsibility for the ongoing war ex-tends to all the civilians who are involved in the politics of waging it,including the leaders and activists of the political parties that supportit. A massive bombing of civilian targets would be necessary in order toeliminate all culpable civilians.

Thus, the war convention immunizes all civilians, by making everysoldier (rather than merely threat-posing soldiers) liable to killing. In-deed, I suggest that it is illegal to kill the engaged, threat-posing civiliansbecause killing non-threat-posing soldiers is legal. There is, in otherwords, a deep relation between these aspects of the convention: thepolitical leaders regain their immunity to killing, thanks to the legalpermission to target soldiers—including those who do not pose a threat.I shall argue that this rule is accepted because (a) soldiers and civiliansbelieve it to be fair, (b) they truly and justifiably believe that it is mutuallybeneficial that it would be commonly followed, and (c) they truly andjustifiably believe that, since others know that it is mutually beneficialthat this rule would be commonly followed, this rule is, in fact, com-monly followed.

Why is it in the interest of the soldiers to immunize engaged, threat-posing civilians? The answer is based on the above factual observation:by doing so, the warfare is removed from the cities to the battlefield—and thereby the chance that helpless innocent civilians will be killeddiminishes. Thanks to this convention, the soldiers’ family members aresafer, released soldiers would have safer places to return to, andwounded soldiers would have protected healing spaces.

But why is it necessary to legalize the killing of nonthreateningsoldiers? Why not just agree on immunizing the threat-posing citizens?Answer: the basic aim of a just war is the elimination of an unjust threatposed by corrupt leaders. Leaders can pose such a threat by virtue ofthe fact that the military forces are subject to their direct control. Hence,an elimination of the threat is possible in one of two ways: either bydisabling the leaders (killing them, crippling them, etc.) or by disarmingthem, that is, defeating the military forces, which are subject to theirdirect control. As noted, most theories of self-defense allow eliminatingthe unjust threat by killing the political leaders who cause it and prohibitdoing so by killing totally innocent soldiers. Yet, ex ante, all sides preferto avoid the first line of action. So we are left with only one way toeliminate the threat posed by corrupt leaders—namely, disarming them.Innocent soldiers, realizing that this is a way of waging war that is overallmuch less destructive than most realistic alternatives, accept the rule


that denies their right not to be killed by the enemy, even if they areinnocent.37

IV. CONCLUSION

I have tried to dissolve the suspicion that the war convention representsa Hobbesian (rather than Lockean) view of self-defense. I argued thatthe initial aggression in the typical war is a crime, but, as Walzer stresses,it is a crime of the political leader. Once soldiers are pushed into a war,an egalitarian just war theory—built on Thomson’s conception of self-defense and on an efficient and fair social norm regarding the protec-tion of the rights to life in the anarchical society of states—is a morallysolid framework for a war ethic.

37. Consider a rule that commands only partial immunization of civilians. Supposethat if it is commonly followed, it would produce the benefit that full immunizationproduces. Even then, the current, cruder convention is justified if the chances that therule that commands partial immunization won’t be commonly followed are sufficientlygreat. Indeed, the simplicity of the current convention seems to be an essential featureof its moral standing. Note further that there might be particular cases in which theinterest of all soldiers is best served if the just war would be aimed at disabling the leaderof the aggressive state that causes it. Yet, there can be no commonly followed rule thatcommands such a practice. For, most soldiers are likely to believe that the war they fightis just. Hence, generally, if rules of conduct in war would depend on the justice of thecause of the war, wars won’t be fought in accordance with any rules. This general claimcalls for a detailed argument, which I hope to provide elsewhere.

a defense of the traditional war convention* · 2015. 2. 22. · no less controversial) conception...

Documents